Severe Morning Sickness Linked to a Surprising Drop in a Key Cellular Stress Protein

Hyperemesis gravidarum, the most severe form of nausea and vomiting in pregnancy, has long been dismissed as an extreme version of ordinary morning sickness. Yet the condition can devastate lives, driving women to lose weight, become dehydrated, and in some cases terminate wanted pregnancies. Now a new study published in Reproductive Sciences has turned its attention to one of the most fundamental stress-sensing systems inside human cells, asking whether the endoplasmic reticulum, the cellular organelle that folds and quality-checks nearly every secreted protein in the body, behaves differently in women suffering from the disorder. The answer, according to a research team based in Konya and Malatya, Türkiye, is a nuanced and unexpected one: a key downstream signal of cellular stress is actually lower in affected women, not higher.

The endoplasmic reticulum is far more than a passive protein factory. When misfolded proteins accumulate inside it, because of starvation, inflammation, oxidative damage, or hormonal upheaval, the organelle triggers a coordinated rescue program known as the unfolded protein response. Three sensor proteins embedded in the ER membrane police this quality-control checkpoint, and one of them, PERK, is the focus of the new work. When PERK detects trouble, it phosphorylates the translation initiation factor eIF2alpha, which globally damps protein synthesis to give the cell breathing room, while selectively upregulating activating transcription factor 4, or ATF-4. ATF-4 then reprograms the cell toward amino acid import, antioxidant defense, and autophagy. If the stress persists and the adaptive response fails, ATF-4 can instead push the cell toward apoptosis, which is why the PERK–ATF-4 axis sits at the crossroads between adaptation and damage in conditions ranging from diabetes to preeclampsia.

Pregnancy is itself a physiological stress test for this pathway. The placenta is a site of intense protein synthesis, low oxygen tension in early gestation, and brisk inflammatory signaling, and previous studies have documented heightened ER stress markers in decidual and placental tissue from pregnancies complicated by fetal growth restriction and preeclampsia. The same research group had previously examined ER stress parameters in gestational diabetes, finding measurable alterations in circulating stress markers. That body of work made a natural question irresistible: if the unfolded protein response is perturbed in other pregnancy complications, does it also leave a fingerprint in the blood of women with hyperemesis gravidarum, a condition whose biology remains stubbornly incomplete despite recent genetic breakthroughs linking it to the hunger hormone GDF15?

To find out, the team led by Bahriye Oznur Oguçbilek and Sevcan Sarıkaya designed a prospective case-control study conducted between August and November 2024. They enrolled 85 pregnant women diagnosed with hyperemesis gravidarum within the first 16 weeks of gestation, alongside 85 healthy pregnant controls matched for early pregnancy. The researchers measured circulating levels of PERK and ATF-4 in serum using enzyme-linked immunosorbent assays, then interrogated the data with group comparisons, Spearman correlation analyses, and receiver operating characteristic analysis to see whether either protein could serve as a diagnostic marker. Ethical approval came from the Karatay University Scientific Research Ethics Committee, and all participants provided written informed consent before enrollment.

The headline result was a significant reduction, not an elevation, of serum ATF-4 in the hyperemesis group. Women with the condition had a median ATF-4 concentration of 4.55 nanograms per milliliter, with an interquartile range spanning 0.82 to 42.86, compared with 6.16 nanograms per milliliter, ranging from 2.23 to 29.27, in healthy controls, a difference that reached statistical significance at p equal to 0.002. PERK, by contrast, was essentially indistinguishable between the two groups, with medians of 3.73 and 3.29 nanograms per milliliter and a p value of 0.965. That dissociation is biologically intriguing. If the upstream sensor is unchanged while its principal downstream transcriptional effector is depleted, it suggests that whatever is happening in these women is not a straightforward amplification of canonical ER stress signaling, but something more selective, perhaps involving post-translational regulation, altered clearance, or consumption of ATF-4 in tissues rather than a simple overactivation of the pathway.

Equally striking was the strength of the relationship between the two proteins. ATF-4 and PERK levels correlated strongly and positively in every analysis the team performed: the Spearman correlation coefficient was 0.837 within the hyperemesis group, 0.709 among controls, and 0.765 across the entire study population, all with p values below 0.001. In other words, the two molecules rise and fall together in the circulation of pregnant women regardless of disease status, reinforcing the idea that they travel as a coupled axis in vivo. But within the hyperemesis group, ATF-4 levels did not differ according to urinary ketone positivity, the standard bedside marker of starvation and fat metabolism that clinicians use to gauge severity, with a p value of 0.282. The circulating stress signature, in short, did not track the most obvious clinical indicator of nutritional deprivation.

Could ATF-4 serve as a diagnostic test? The receiver operating characteristic analysis suggests caution. The area under the curve was 0.638, statistically significant at p equal to 0.001 but well short of the discrimination clinicians would want in a standalone test. At the optimal cutoff, ATF-4 achieved a specificity of 77.6 percent, meaning it rarely mislabeled healthy pregnancies as diseased, but a sensitivity of only 45.9 percent, meaning it missed more than half of true cases. The authors are appropriately measured in their interpretation, describing the discriminatory performance as limited and framing the finding as a hypothesis-generating result rather than a clinical tool. The value of the study lies in what it reveals about mechanism, not in any immediate application at the bedside.

Placed in context, the findings add an unexpected twist to a rapidly evolving field. The dominant molecular story in hyperemesis research over the past several years has centered on GDF15, the placental and appetite-regulating hormone that acts on the brainstem to suppress feeding. Landmark genetic studies published in Nature and Nature Communications identified GDF15 and IGFBP7 as risk loci, and a 2024 Nature paper linked maternal GDF15 levels directly to nausea and vomiting risk, culminating in a multi-ancestry genome-wide association study of severe pregnancy sickness. Meanwhile, oxidative stress has been documented in hyperemesis patients, and systemic inflammation markers have been associated with the condition in recent case-control work. The new data suggest that the unfolded protein response, long implicated in placental pathology, deserves a seat at the same table, even if its circulating signature points downward rather than upward.

Why might ATF-4 be lower rather than higher? The authors themselves do not overclaim an explanation, and the honest answer is that the study cannot distinguish among several possibilities. Chronic starvation and ketosis can suppress integrated stress response signaling in some tissues; the adaptive phase of the response may have resolved by the time of sampling; or circulating levels may simply mirror what is consumed and bound in metabolically active tissues rather than what is produced. It is also worth remembering that ATF-4 is a transcription factor whose biology is governed by context: in the placenta, the same pathway that protects trophoblasts under mild stress can drive apoptosis under severe or prolonged stress, a duality that recent reviews have called the stress paradox of the unfolded protein response in pregnancy. A single serum measurement captures one frame of a dynamic process.

The study’s limitations are those of any cross-sectional case-control design: it captures a snapshot, cannot establish causality, and relies on peripheral blood rather than placental tissue, where the most consequential ER stress biology likely unfolds. The authors call explicitly for molecular and longitudinal investigation of the PERK–ATF-4 axis in hyperemesis gravidarum, and that is the right next step. Serial sampling through pregnancy, tissue-level studies of placental and hepatic stress signaling, and integration with the GDF15 and inflammatory axes could reveal whether the depressed ATF-4 signature is a cause, a consequence, or a bystander of the disease. For the roughly one to three percent of pregnancies affected by hyperemesis gravidarum, a condition the Windsor international consensus definition now recognizes as a distinct disorder with serious consequences for quality of life, any new molecular handle is welcome. This study does not yet change clinical practice, but it opens a genuinely novel window onto a condition that has, for too long, been understood mostly by what it is not.

Subject of Research: Endoplasmic reticulum stress markers in pregnant women with hyperemesis gravidarum

Article Title: Investigation of Endoplasmic Reticulum Stress in Pregnant Women with Hyperemesis Gravidarum

Article References: Oguçbilek, B. O., Kuzu, K., Akat, F., Almaghrebi, E., Vatansev, H., & Sarıkaya, S. (2026). Investigation of Endoplasmic Reticulum Stress in Pregnant Women with Hyperemesis Gravidarum. Reproductive Sciences. https://doi.org/10.1007/s43032-026-02233-1

Image Credits: AI Generated

DOI: 10.1007/s43032-026-02233-1

Keywords: hyperemesis gravidarum, endoplasmic reticulum stress, ATF-4, PERK, unfolded protein response, pregnancy, placenta, GDF15, case-control study, ELISA, morning sickness, Reproductive Sciences